Although a few published studies have investigated reproductive IVF/ICSI outcomes among women with PCOS, oocyte morphologic quality for the various PCOS phenotypes has not previously been described. Indeed, previous research by our group found that oocyte quality was similar in a PCOM group and a control women; however, the study’s PCOM group contained a mixture of PCOM-only women and PCOS women [36]. It is well established that PCOM is a distinct phenotype and is not the same as PCOS [30]. We therefore decided to investigate oocyte morphology among the different PCOS phenotypes. To the best of our knowledge, the present study is the first to have addressed this topic. Our study population was homogenous and met strict inclusion criteria - including ICSI for male infertility only. All cases with potential female infertility other than PCOS were excluded.
Importantly, we did not detect any significant differences in oocyte morphologic quality or embryo quality between the PCOS phenotypes A, C and D. This unexpected observation challenges a number of preconceived ideas. Firstly, some researchers have reported that relative to healthy women in IVF and ICSI programs, phenotypes A and B (considered to the most severe, with an increased risk of comorbidities such as hyperinsulinism and metabolic disorder) are associated with a lower pregnancy rate [18] and a greater risk of adverse outcomes in pregnancy [16]. It has been suggested that these poor outcomes are related to low oocyte quality or competence. Secondly, the cardinal features of PCOS (i.e. OA, HA, and PCOM) are suspected to have an adverse impact (independently or combined) on oocyte quality [15].Indeed, androgens are involved in folliculogenesis, and a hyperandrogenic environment leads to abnormal folliculogenesis, prematurely activated follicles, mitochondrial abnormalities, and failure of meiosis progression to MII. Furthermore, HA is known to induce premature luteinization of the granulosa cells, which prevents them from progressing to physiological atresia. It has been shown that OA is associated with an alteration in the insulin growth factor (IGF) pathway, which is involved in embryonic development and blastocyst formation [12]. However, the true extent of the IGF pathway’s involvement in the pathogenesis of PCOS is still unknown. Hence, the impact of oligo-anovulation on oocyte quality and reproductive outcomes (with the exception of the pregnancy rate per cycle) has yet to be determined [15]. Likewise, the impact of HA on oocyte quality is subject to debate. Some researchers have highlighted a negative effect of androgens on oocyte maturity in animal models [44], and others have demonstrated the androgens’ fundamental role in early folliculogenesis and the pre-ovulatory follicular stages [45]. The elevated androgen levels in PCOS might lead to excess AMH secretion, which in turn is involved in so-called “follicular arrest” in the ovaries [34, 46]. In contrast, Gaddas and colleagues (2016) did not find any negative impact of biochemical HA on conventional IVF or ICSI outcomes in women with PCOS [17]. Furthermore, a study by Palomba and colleagues (2010) did not show any significant effects of clinically defined HA and PCOM (PCOS phenotype C) [16], which is in line with our present results. Lastly, our comparison of oocyte quality in PCOS, PCOM-only and controls gave much the same results as the only two other studies to have evaluated morphologic abnormalities of the oocytes [10, 36].
Our present results confirmed the above-mentioned observation for clinical and endocrine features in PCOS phenotype groups. Firstly, we did not find any differences in the woman’s BMI between the three A, C and D phenotypes - as also reported in a recent case-control study [39]. Secondly, the PCOS phenotype A population exhibited the highest AMH levels, which is in line with previous reports [18, 19, 39, 40]. Furthermore, the severity of OA appeared to be correlated with an elevated serum AMH level [41, 42]. We discovered that serum delta-4-androstenedione levels were higher for PCOS phenotype A than for phenotype D, and that LH and AMH levels were higher for phenotype A than for phenotypes C and D. These results are also in line with another recent report [19].
It is acknowledged that oocyte morphology is an easily assessed, non-invasive marker of oocyte quality. Some oocyte abnormalities (i.e. a large perivitelline space, the presence of cytoplasmic vacuoles, or an abnormal shape) are known to be associated with poor reproductive outcomes [33, 37, 45–47]. In contrast, some researchers did not find any differences in these morphological parameters [34, 48, 49]. Likewise, our study did not evidence an adverse relationship between individual oocyte morphologic abnormalities and the ICSI outcomes. Consequently, the oocyte morphology’s predictive value is still subject to debate [50, 51]. We therefore assessed oocyte quality with regard to not only the extra- and intra-cytoplasmic abnormalities described above but also two scoring systems. Each oocyte abnormality is considered separately in the AOQI, [36], whereas the presence of specific oocyte features are computed in the MOMS. In the present study, we did not detect any differences between the PCOS phenotypes A, C and D with regard to the AOQI or the MOMS.
In line with previous reports of similar embryo quality in PCOS and non-PCOS patients [8, 43, 54, 55], we did not observed interphenotype differences in these variables or, indeed, in reproductive outcomes. This contrast with recent reports in which PCOS A and B were associated with a lower clinical pregnancy rate [18] and PCOS phenotypes with HA were associated with a lower cumulative live birth rates, when compared with normo-androgenic counterparts [56]. This discrepancy might be due to our strict inclusion criteria and thus our small study population. However, our study was not designed or powered to assess ICSI outcomes.
Our study had a number of limitations. Firstly, we cannot draw any conclusions with regard to the PCOS phenotype B because we used the serum AMH level as a surrogate marker of PCOM. In fact, our patients with PCOS patients were diagnosed according to the revised Rotterdam criteria [3, 25, 33]. Secondly, we used a serum AMH threshold concentration > 35 pmol/l for the PCOM group in our study [27, 28] This threshold was established with an assay kit that is no longer commercially available [57]. For consistency, the inclusion period for the PCOS, PCOM-only and control groups was restricted to the time when we used the AMH assay; this also explains why our study population was small. Some researchers consider that the serum AMH level is a more reproducible, more sensitive variable than the FNPO [16, 27, 28, 41, 58, 59]. The FNPO threshold of 12 in the 2003 Rotterdam criteria is outdated and leads to overestimation of the prevalence of PCOM in the general population [60–61]. Consequently, we decided to use the FNPO threshold of 19 reported in a cluster analysis [27]. This value is also close to that given in international evidence-based guidelines for the assessment and management of PCOS from 2018 [4]. An AMH assay and pelvic ultrasound should be used together to determine the presence of PCOM and thus correct a false negative for one or the other. Hence, their combined use causes the phenotype B to almost completely disappear [28].
Secondly, our study’s retrospective, single-center design was associated with inherent bias. Nonetheless, the single-center design enabled us to generate novel data in a homogeneous patient population. All the women received standardized care, and the proportion of missing data was very low - constituting key study strengths. Furthermore, we used very strict inclusion and non-inclusion criteria, notably with regard to any differential diagnoses; this also explains why the study population was relatively small. Furthermore, we used Poisson and logistic regressions to assess the effect of potential confounding factors. One limitation of this study was represented by the technical improvement of ultrasound over the past ten years which can impact the antral follicle assessment, in particular regarding small follicles. Nevertheless, our FNPO values were all obtained with transvaginal ultrasound using a 5–9 mHz probe, based on a routinely standardized protocol in order to minimize the impact of technology improvement on follicle counting.
In summary, our results showed that PCOS A, C and D phenotypes did not differ significantly with regard to oocyte morphologic quality and thus suggest that no single PCOS phenotype is associated with poor quality. Our results also suggest that the phenotype A (considered to be the most severe phenotype) is not associated with especially poor reproductive outcomes. Further studies are required to confirm these findings and characterize the underlying mechanisms.